Related
The lovely thing about having a Linux Kernel is that we can do a lot to optimize it.
I figured it would be a great idea to set up a thread where we can bring all these optimizations into one place.
Mind you, I'm no expert by any means...
For my (possibly useless) contribution:
When we compile the kernel to the G1, do we compile it with the general Arm eabi? If so, is that made for ALL ARM cpu's? We have an ARM6 in our G1's. If we set it up to optimize to ARM6, can we set it up to use the CPSID/CPSIE instruction instead of the SWP instruction? Are the drivers open source? Can we recompile them to do the same also? To learn more about why, read this:
http://forum.xda-developers.com/showthread.php?t=516223
That's where I got the idea from.
SWP is a deprecated opcode in armv6 that decreases performance due to the way it locks the bus.
Click to expand...
Click to collapse
Anyhow, whatever you guys post here, I will add to this post. Also, to the people to compile the kernels, can you answer my questions?
answers to some of your questions.
hi i usually don't post here since i am not a developer but in this case i think i can answer some of your questions (since i sometime make custom kernels for my laptop). kernel optimazation is used to removed excessive drivers and services from the kernel that you don't need that slows down your computer or in this case phone. i believe htc already does this when they configured android for the g1. most cell phone manufactures already do this to ensure that there products processor use energry efficiently and and run as smoothly as possible.
When we compile the kernel to the G1, do we compile it with the general Arm eabi? If so, is that made for ALL ARM cpu's? We have an ARM6 in our G1's
Click to expand...
Click to collapse
yes, arm eabi is made for all general arm cpu's i believe you can complie it for arm 6 specifically though. how do you know that htc hasn't already optimize the kernel for that specific processor? also if they didn't you wouldn't really see any noticible speed improvements by compliling the kernel for this specific processor.
Are the drivers open source?
Click to expand...
Click to collapse
no, futhermore they don't offer the drivers for the g1 except execpt for the usb driver so your computer can recognise the g1 as mass storage device.
If we set it up to optimize to ARM6, can we set it up to use the CPSID/CPSIE instruction instead of the SWP instruction?
Click to expand...
Click to collapse
i don't know sorry. thats all i know i'm really not a kernel, coding expert i hope this helps in your task.
bananajama said:
no, futhermore they don't offer the drivers for the g1 except execpt for the usb driver so your computer can recognise the g1 as mass storage device.
Click to expand...
Click to collapse
I think you meant to say "ADB interface" instead of "mass storage device."
t1n0m3n said:
I think you meant to say "ADB interface" instead of "mass storage device."
Click to expand...
Click to collapse
The system type is set to "Quaalcom MSM", so I would imagine that it's optimizing for arm6l.
cyanogen said:
The system type is set to "Quaalcom MSM", so I would imagine that it's optimizing for arm6l.
Click to expand...
Click to collapse
Sounds like it. Well, other than that, are there any other obtimizations anyone else can think of?
there are few things in linux that you can play with to see if you get any performance increase. It's mainly in memory management. run ls on /proc/sys/vm. you can play with the page cache but I doubt that you'll get much perf increase from it since, since most of the things we can play with are for write back cache. If we have access to the dalvikvm we may be able to do more optimization. The one thing that I've always wanted to be able to do is to pin Home in memory, and not allow garbage collector to kill it, so when we exit out of memory intensive apps, like browser, gallery, etc, we do not have to wait for Home to reload. If we have more control over garbage collection we may be able to do more, but I don't know much about the dalvikvm at all.
taking a quick look at the vm. The only nob that may have any effect on android may be vfs_cache_pressure. theoretically setting to 200 would push fs cache out quicker and give more memory to the vm. default is set to 100. I set it to 200 and 300 but didn't notice any difference.
knaries2000 said:
there are few things in linux that you can play with to see if you get any performance increase. It's mainly in memory management. run ls on /proc/sys/vm. you can play with the page cache but I doubt that you'll get much perf increase from it since, since most of the things we can play with are for write back cache. If we have access to the dalvikvm we may be able to do more optimization. The one thing that I've always wanted to be able to do is to pin Home in memory, and not allow garbage collector to kill it, so when we exit out of memory intensive apps, like browser, gallery, etc, we do not have to wait for Home to reload. If we have more control over garbage collection we may be able to do more, but I don't know much about the dalvikvm at all.
Click to expand...
Click to collapse
dxtop, which is a home replacement, does exactly that. It keeps itself in memory so going back to it is much faster than the default home application. And in low memory situations, it tries to free up as much resources as possible while still staying in memory.
ok its simple we all know the benefits of more ram. we all know that android slices up the ram into static addresses within the kernel. i would like to change that as we are in 2009 almost 2010 and we still rely on static ram allocation for the camera, gpu, sensors, radio. the g1 has 192mb of ram. yet we only get 97 for programs to use. i would like to place the camera, gpu, etc into dynamic addresses that we can change within the os without flashing a new kernel. a service, kernel code, module, or any other method could do this. we just have to make it happen. we could do it three ways i see.
1. a system managed ram allocation using a service to detect the percentage used and increase ram on the go.
2. a kernel script or service that only activates and allocates ram when needed. such as once a 3d program is loaded the 8mb magically comes back.
3. a virtual device ram allocation service that makes the device think it has the ram all along, but only actually gets it when it needs it. kinda like the way apps2sd works.
4.(by IConrad01) run perl scripts through ase and use ksplice to patch the gpu1 memory in kernel. one script on and one for off.
any other methods to implement this proper are very welcome and encouraged.
the idea behind this is a proper way to get back more ram for our programs. the idea is simple when 3d is not in use we can used the 8mb it would have used for something else. this applys to everything that has its own ram. camera, etc. say we did this and only got 120 free. thats almost 25% more ram. plus we could go the extra step and let the user choose how much each thing gets(gpu settings 4mb,8mb,12mb. camera on, off. etc) so when it needs it thats what it gets. great if you like/dislike 3d games.
on another note this is the kind of work that gets you noticed. the kind that gets you hired or gets you respect. google would be proud and might implement it and you and i know it. i hope we are successful and google uses it mainstream.
so lets debate a way/ways to do this
provide alpha tests of different ways to do this
and finally lets post a beta so everyone can enjoy.
atm ksplice with perl scripts is where we are going will update as this progresses.
Doesnt Jacs update to his hero rom have this, im probably mistaken
jad011 said:
Doesnt Jacs update to his hero rom have this, im probably mistaken
Click to expand...
Click to collapse
No. The current implementation of this basically cuts the graphics memory in half.
This would dynamically allocate the memory as needed. Much more efficient way of doing it.
he has a static change that takes the ram away from the gpu by a kernel edit. this is to have both more ram but still give the gpu what it needs to perform properly but only when it needs it or you chose to give it.
Are you referring to RAM or storage space? You say RAM in the title but then it sounds like you're talking about storage space before making me think it's RAM again later on. I'm confused :/ I'd really like additional free RAM, anything that helps reduce swapping to SD card is good but storage space I have plenty of.
all of this would apply to ram only. swap and compcache could use the same idea to dynamically allocate swap but that is still a cache for programs. sorry if its confusing. simple way to explain. i want all the ram i can get but still use everything the phone has properly, like games and the camera.
Gotcha and understood. I see so many people say RAM and memory when they're referring to storage space sometimes that I want to scream but extra RAM on the G1 would be more than welcome. Hopefully others with the brains to make it happen can do it
actually anyone with a basic knowledge can help. mostly there are a couple things to developing a good idea.
first the concept.
then ideas on how to do it.
next a map or guide to what does what.
coding different parts
putting them together
debug and test
even if all you do is photoshop the ui or provide ideas it helps. once some different people get aboard this should be working sooner than you think.
Of the three options listed, the kernel script (via ksplice) seems like it would be the shortest route to functionality, but frankly it also seems a bit... well, hackish -- when compared to the virtual device allocation scheme. It would take a hefty re-write of the kernel, I suspect, to accomplish; but being able to resize various devices on the fly would be a good thing. Shrinking CompCache based on real RAM usage for example would probably speed up the phone under light loads.
Just thinking in text here.
personal i think its going to start simple(script or root app) and go straight to gpu shared on or off. choose and reboot. however i do not want to stop there like winmo did. someone could create a short kernel patch that checks for a service or script and gets value or defaults to standard method. then we would have to come up with how to manage the switch(realtime vs. reboot) ideally we would remap the memory from scratch and rebuild the memory management services to dynamically allocate the ram for all devices(128+ ram) as needed. we have a start, we know where and how to turn the gpu1 shared on or off. now lets make it more than a hack and work on how to make our phones better. btw this is not a waste of time. this extends to other phones and the future way to set up our phones from default.
jokersax11 said:
personal i think its going to start simple(script or root app) and go straight to gpu shared on or off. choose and reboot. however i do not want to stop there like winmo did. someone could create a short kernel patch that checks for a service or script and gets value or defaults to standard method. then we would have to come up with how to manage the switch(realtime vs. reboot) ideally we would remap the memory from scratch and rebuild the memory management services to dynamically allocate the ram for all devices(128+ ram) as needed. we have a start, we know where and how to turn the gpu1 shared on or off. now lets make it more than a hack and work on how to make our phones better. btw this is not a waste of time. this extends to other phones and the future way to set up our phones from default.
Click to expand...
Click to collapse
The only problem with this is when the ram is divided, everything in it is cleared, thats why it is set at boot in the kernel. Think of it like repartitioning your SD. Now it would probably be possible have everything mapped to ONE big RAM table & then build something similar to a scheduler like BFS for RAM that would allocate it as necessary. This is messy though.
Just my $.02
well yes but it is possible to create shared memory through c++ http://www.go4expert.com/forums/showthread.php?t=8461
however i think the vm heaps has either a 6 or 16mb limit. anyways the service would basically do a memory patch over the kernel in realtime(like overclock) telling the gpu the memory is there all the time. meanwhile it allocates dynamic shared memory from the application memory for the gpu to read and write. meanwhile an app tied to a script would set the parameters and settings.
Okay -- I know ksplice works with perl; and I know that we can run perl scripts using the ASE .apk.
So... it seems to me that it /should/ be possible to use a patch that reverts the patched kernel back to the previous state, and a patch that switches the original state to the GPU1=0 MB state. From there, we could set up two scripts; one for GPU1-off and one for GPU0-on. Each would simply be an invocation of ksplice to input the specific patch (which would have to be stored on the sd card. [it would be safer if in the ext partition]).
We'd have to set up a secondary RAM device, rather than just adding it to the standard RAM pool... something like establishing it as a /dev/block/shm.
The point of all the above is that when you click on the GPU1-off script, it "reclaims" that RAM and gives you a device that's standard RAM. When you click on the GPU1-on script, it disables that device (forcing the reload of all programs using that RAM) and re-enables the GPU1 device.
Even better yet, the Android Scripting Environment program, much like GScript, supports desktop icons.
So -- very initially speaking, assuming we can get ksplice working on Android (which just requires a static binary, IIRC), we can get a simple pair of icons; one of which frees up the GPU1's RAM; and the other of which re-enables the GPU1.
Since the only times it's going to be "called up" are when a game is actively being played, or something similar, this should be a mostly-satisfactory route to begin with, before we can get into the whole scheduled/scaled script running as a daemon in the background.
Once we get /there/, then we can start looking at true virtual device allocations and shared memory and the like.
Although, I do have to say that I'm not entirely sure how comfortable I am constantly repatching the kernel. Definitely, we're going to want to recommend that people back up their original, unpatched, boot.img files!
IConrad01 i added your method to first post as it was a great start to this project. if you want me to rephrase just post what you want and ill change it. the only issue i might see is what geniusdog254 said it might unallocate the wrong thing and create a reboot. but some clever coding should prevent this.
Nah, that's all good.
There's an easy way to prevent it from becoming a reboot-able issue. Enable the GPU-space device as "swap". This could be done one of two ways; tell the phone that the new 8 MB RAM "device" is to be considered the /first/ swap device to be used, and just leave it "as is". The r/w times would be for true RAM. The other option assumes we have the latest CompCache (0.6+) -- as it allows for multiple instances of CompCache. The theme here should be obvious.
Either way, the result is that only swappable pagefiles would be lost when the device is turned back into GPU memory. No system-vital memory would get tossed. Seems simple enough to accomplish, perhaps. But then again, I Am Not A Programmer.
ok so what are the alternatives to ksplice for the swap device change. since we could make it swap/compcache anyways is it possible to change on the fly like the way overclocking widget does?
Ksplice allows you to make kernel updates and have them implement themselves without having to reboot the device. The swap device stuff would have to be implemented on a script-level. Two different things: I.e.; the script invokes ksplice to install a specific patch, which causes the kernel to recognize the newly freed-up RAM as a specific device (I.e.; such as /dev/ram1 as opposed to /dev/ram0). I am not sure I'm saying/thinking of this properly here: I'm operating on the edge of what I know about Linux.
Anyhow -- the trick is, when we get the kernel to disable the GPU's memory, we enable a new RAM-disc swap partition with its swappiness set to the maximum rating so it picks up everything first. Since it is /actually RAM/, it'll perform like real RAM. We're only 'declaring' it as swap in order to prevent forced reboots when turning it off.
The above wouldn't just be a ksplice thing; ksplice is just to implement daproy's kernel patch and thus free up the devoted memory to the GPU1. This memory would then be allocated to a new device (thus preventing the RAM-dump forced reboot) and /that/ would be set to swap status to ensure no "mission-critical" files go on it.
Then, once we get that built up, we just build it in "reverse" -- turn off swap (dumping its info back to the system, where CompCache / linux-swap would pick up the load) and re-enabling the GPU1.
Once we get /those/ built, we can then create a daemon script, include an invocation of it into the latest userinit.sh, and have that script basically monitor GPU0's load levels. If they go above a certain threshold (scaling, just like with the CPU), it would invoke the second script above, patching the kernel and re-enabling GPU1. This would then monitor the GPU, and once //its// memory load falls back down to where GPU0 can handle it all, turns GPU1 off and frees that RAM back up for individual use.
I can already foresee a complication where turning GPU1 off "on the fly" would cause a reload of all graphics, but... considering how rare it is to even pull a load above what GPU0 can handle in ordinary use, this isn't much of an issue. It'd take some tweaking and beta-testing (which I'm pretty much set to be a tester of at this point, apparently ) but that would be about it.
Now, //ideally// we could create an alternative solution which wouldn't rely on constantly patching the kernel. As I said; that can get... well, "messy".
I just don't know enough to know what that alternative path would be. Virtual shared memory space seems like it would be a good idea. An alternative solution could be something along the lines of enabling GPU1 at all times and simply having it call to a swapfile which is located in /tmp (I.e.; a non-compressing swap-file which is actually located in RAM). Said swapfile could be resized freely according to need and would thus be no issue at all.
Frankly, I can already see one reason why that would be a superior solution to all the ksplice shenanigans above: said swapfile could be made to be /larger/ than 8 MB, and thus actually provide for an increased performance due to higher available memory for the graphics side of things. We'd want some way to make sure that the swapfile itself wouldn't get pages dumped onto an actual swap partition. This seems like a much more elegant solution than the ksplice routine. I simply have no idea if it's even feasible. The swapfile would have to be mounted into /tmp (or it's Android equivalent) //at the time the GPU1 was enabled//. Is that even /possible/ at boot? Could we simply enable the device very late in the boot stage?
Quickly! Someone with /actual programming expertise/, rescue me from my aimless mental tinkering!
Since it seems there's little activity in this thread, I thought I'd share:
There is a debian armel package for ksplice.
( http://packages.debian.org/squeeze/ksplice )
This is ARM-compiled. We would simply need to cobble together a script which invokes ksplice properly. I myself know nothing of how this would be done.
Usage example of ksplice here:
http://www.ksplice.com/example-update
Looks pretty simple. You just need a working kernel build environment. Anyone got one of those laying around? I have wanted to do that for awhile but am too lazy to set up a linux VM, lol.
Hell yeah!!
Im not much good for this project other then willingness to try betas and give feedback, but I'm excited to see where this goes.. It would be great to reclaim the ram being used from the system when you aren't using anything that needs it and use it for what you are using. Has anyone contacted one of the devs that have been modifying the kernels and ask them if they could try building it with the ksplice functionality?? I would think one of them would be happy to build it so you can start playing with it. Wish I could help more on it, but this is a little above my head, but I will be following this thread and learning what I can as i go.. Good Luck!!
Attention!
If you're looking for scripts, here it is:
[CWM][SCRIPTS][TWEAKS] Thunderbolt!
Script Reviews are here:
Script Reviews
Introduction
I've been meaning to share this with the whole Android community. It seems unfair that only the i9000 folks have access to the vast research that I've done so far. Hence I'm sharing this in the general Android thread in hopes that it'll benefit everyone in the long run
A lot of people often asked about how Android really works and the optimizations can be made to their Android to make it perform better in terms of:
- battery life
- raw performance
- GUI smoothness
With Android based on Linux, and with the experience I have with Linux/Unix (I deal with Unix/Linux on a daily basis for my work) I managed to find some tweaks that we can do to optimize our phones.
I also hope that people will experiment with the tweaks posted here and feedback to me if there are certain ways that I'm doing incorrectly or there are better ways than what I've posted here.
Script tweaks
There are a lot of scripts in XDA that specifically uses the optimizations (Linux or Android based) that I will explain below to actually increase performance/battery life. However, you need to know what those scripts do before using it.
I've done some reviews of the script here:
Script Reviews by Pikachu01
There are good scripts that are properly tested and made sure that it works. It's a script by Zacharias maladroit, the kernel dev who makes the platypus kernel for CM7 and snail kernel for i9000. I modded it to be compatible with i9000 and it'll probably do more than some of the scripts I reviewed
From there, I've also taken the liberty of modifying it some more based on my knowledge and research I've done. It's called ThunderBolt!:
LINK
Now onwards to what actually goes on beneath the hood.
Android Governors
Some custom kernels provide customized governors for you to choose. These are currently known:
- Ondemand (CPU scales to the highest frequency immediately after a certain CPU threshold)
- Conservative (CPU scales to the highest or 1 scale lower than that after a certain threshold gradually and scales down if CPU is below a certain threshold
- OndemandX (Has sleep codes that scales down the CPU when device is asleep. Threshold is set to scale slower)
OndemandX
OndemandX has a suspend frequency that it maintains when the phone is sleeping. It does have a known issue of waking up too slowly when a call comes in.
Conservative
Conservative is used when you would like the CPU to scale to the maximum through each frequency slowly before reaching the maximum. It saves more battery at the expense of smoothness. There are a few tunables that can be configured in Conservative governor, namely the freq_step and both the up and down thresholds.
freq_step is a tunable that determines how much frequency (a percentage of the maximum frequency of your device) to increase at each sampling rate. e.g. you have 5 as freq_step and your maximum frequency is 1GHz. At each sampling rate, it'll increase by 50MHz. If your current frequency is 100MHz, it'll increase to 150MHz at the next sampling. What if your device doesn't support 150MHz? This is done by splitting time between 100MHz and 200MHz (assuming your next frequency is 200MHz) so that in average, your phone is performing at 150MHz.
The up_threshold is how much threshold to wait for before increasing the base frequency to the next freq_step. I.e. if it is 80, the CPU will wait until it is 80% loaded before increasing the frequency. The down threshold is the opposite of up_threshold, which is the threshold to wait until the CPU downscales the frequency to the next frequency step.
Ondemand
Ondemand is a governor that basically a simpler version of conservative. One difference is that it scales to the maximum frequency after load has exceeded a threshold. This is usually 80. After that, it looks at the load again when it's at the max frequency. If it's below 80 (example), it'll scale down one frequency step. The frequency step mentioned here is the available frequency step. I.e. if your device has frequencies of 100,200,400,800 to 1GHz then it'll scale down from 1GHZ to 800, 400 then 200 and 100.
One optimization I have for Ondemand for i9000 TalonDev is that I tuned it to be 80 rather than 65. I have not felt any slowdowns from using this setting and it saves a tiny bit more battery. You can experiment on this value however you want, but bear in mind that I'm not responsible for any slowdowns or damage that is caused by it. Read up a bit more on thresholds to know more (Google it).
I/O scheduler tweaks
Another tweak that can be done is the I/O scheduler tweak. Some kernels come with a few standard I/O schedulers while others have extra schedulers that you can choose from. I'll explain the few that I know and if you have more information on the schedulers, you can post it here for everyone to discuss/share
CFQ - Completely fair. Most will come with this, but this is not optimized for android. Some kernels do optimize it and use it.
Noop - Simple and least overhead of all I/O schedulers. Produces the best outcome for some cases and some do swear by it. It's downside is that it is easily bogged down by high I/O transactions.
Deadline - Optimized for mobile-like devices like Android. Also, some do swear by this. If its writes_starve sysfs is tweaked to be fair, it's the same as SIO.
SIO - It's a fair deadline scheduler. It's the best scheduler. Nuff said.
VR - A newer I/O scheduler that places penalty on reverse seeks. Not for flash drives, though the R penalty can be tweaked to be a multiplier of 1. Even if this is done, it has a high overhead.. Read more here: http://www.gelato.unsw.edu.au/IA64wiki/IOScheduling/VRscheduler
BFQ - Too high of an overhead, it is optimized for spindle disks.
You can pick and choose from Voltage Control or other UV/OC apps and apply it immediately to feel the difference.
Here's my assessment of all the schedulers that I know:
SIO> NOOP> Deadline > VR > BFQ > CFQ
NOOP is a simple I/O scheduler without overhead that tries to do each I/O transaction as it comes (FIFO). When a group of transactions is detected, it will try to merge it together to make batches of transactions (makes the whole transaction faster). NOOP doesn't have starvation detection, hence if an I/O transaction takes a painfully long amount of time, it will still continue to do it rather than switch the CPU into doing something else e.g. GUI interrupts (i.e. scrolling lists, flicking screens). All other schedulers also have the "merge" feature. NOOP is the only one that makes the "merge" feature its only feature.
CFQ is a complex I/O scheduler that tries to determine the address space of the transaction and applies a cost algorithm in that if the address is close together, it will group them up and perform them. It also tries to make the transaction incremental (i.e. reading/writing through address incrementally so that the disk spindle needs to wind down the least in conventional platter hard disks) The problem is, our flash devices have very little delta between reading a far reaching address space (than the one currently being written/read) or a closer one as it doesn't rely on spindle/rotations. Hence, having this costing algorithm adds overhead and slows down the overall transaction. CFQ has a lot of algorithms to make sure each process gets a fair slice of time on I/O transcations. Too much overhead.
Deadline has a starvation detector and is simple enough that it doesn't have all the overhead of doing rotational/costing disks algorithm. However, reads are done 2x more likely than writes as it has a algorithm based on weights in that reads must be done first if both a read and a write is detected. It has a 2:1 ratio of read to write weights coded into the scheduler (that can be tweaked - writes_starved, will include it in the next version of system_tweak). Hence it's not a fair scheduler.
SIO is Simple I/O that tries to implement a NOOP type scheduler that has starvation detection. Hence, long I/O transactions will be preempted and given CPU time only after other faster transactions are completed, guaranteeing smoothness. Also, it doesn't have overhead of calculating costs. Also, it has a fair number of writes to read, guaranteeing that all transactions are equal.
BFQ has a lot of algorithms to make sure each process gets a fair slice of bandwidth (Budget Fair Queueing). Too much overhead.
V(R) tries to make sure that each transaction has a weight associated with it, being R. And if the seek is reversed, the R will be multiplied by a penalty making it less likely to be processed. Not for flash drives as reverse seek in a flash drive is just as fast as a forward seek.
In benchmarking tests, the tests normally consists of testing the time it takes for a contiguous I/O transfer from 1 point to another. NOOP excels at that because it won't let itself get interrupted to perform another I/O task. This would mean that in real life testing, NOOP will let a long arduous task to finish while at the expense of UI functionality (you will get UI lags)
SIO on the other hand will perform badly at benchmarking as it gets pre-empted when the contiguous tasks takes more than 0.5secs (for synchronous tasks as benchmarking tools perform a synchronous I/O task from one point to another) to another more important task like UI interrupt (when you're scrolling) or Kernel interrupt (when your kernel needs to perform garbage collecting or swap memory etc)
The 0.5 secs can be tuned in the SIO tuneables though, but I would think 0.5 secs for synchronous tasks and 5 secs for async tasks is pretty good to maintain a balance between long/short tasks enabling a smoother experience in Android.
LowMemoryKiller
The LowMemoryKiller is a constant debate between more free RAM and more multitasking capabilities as free RAM (more than 60MB free) is actually wasted RAM. The fact is SGS only has a puny amount of RAM left after a few big chunks are allocated to the Graphics, Modem, Sound and Video (and some others that I do not know a lot about). With that knowledge, Samsung decided to give the SGS SL a bigger amount of RAM when they released it ( about ~650MB of RAM read that from somewhere, can't remember).
The LowMemoryKiller is actually a feature in the Android OS for better memory management.
You can read about it more here:
http://forum.xda-developers.com/showpost.php?p=5442369&postcount=1
This is an important feature due to the perennial problem of having low free memory causing lagginess and slowness in launching apps. When you have free memory lingering around the number of 40MB or less, the Android OS just lags like hell.
What this would mean is, you would want to tweak the LMK to not have the situation of it having less than 40MB (or even close to that).
The modern Linux machine in the Android ecosystem relies on a mechanism called Low Memory Killer (LMK) to consistently free up RAM. This is due to Android's internal mechanism of caching apps (and never fully exiting them) when you press the back button. This is to enable faster app switching and provide a seamless experience for apps usage model. Android also, by itself will also constantly look for often used apps to cache them for faster app opening. This will happen even before your system fully boots.
Now, when you mention LMK, the most obvious thoughts that come up are minfrees and Out Of Memory (OOM) groupings. Yes, those two are integral parts when it comes to LMK. The issue here is that no one actually mentioned that there are two LMK systems in Android, that being:
- Linux LMK
- Android Dalvik VM LMK
Both are separate entities that kills/removes app/dalvik cache from RAM when RAM reaches a certain level.
What confuses most people (including me) is that the OOM groupings of both mechanisms have the same names being (Android 2.3 based):
LMK App Categories
FOREGROUND_APP: Your apps in the foreground, being used currently, interfacing with the user.
VISIBLE_APP: Visible app that is still viewable by the user, but not interacting with the user, example could be apps that reside in the status bar, and giving live information i.e. Os monitor graphs.
PERCEPTIBLE_APP: App that the user can still perceive i.e. Music app that is playing music.
HEAVY_WEIGHT_APP: RAM heavy apps that are not being interacted with, but will be a pain to load if its information is cleared in the memory.
SECONDARY_SERVER: App that acts as a secondary server. Not sure what this really means, but client-server thingy but secondary? Could be something that syncs with an app, but is not syncing currently?
BACKUP_APP: App that is currently making a backup (like Titanium backup?)
HOME_APP: Your launcher.
HIDDEN_APP: An exited app that still significant residual memory in the RAM. Exited only some time ago, or is constantly used by the user.
EMPTY_APP: An exited app that only has small remnants of residual memory in RAM. App that is exited quite some time ago.
Take note that this is my understanding of the OOM groupings. If there are mistakes, please correct me
Tuning LMK
Both the LMK and Dalvik VM has this groupings and they can be tunable by using prop and lowmemorykiller/parameters/adj respectively.
One errata about this is that the Dalvik LMK has one extra tunable which is CONTENT_PROVIDER, which affects widgets that are not currently refreshing or displaying information. The ADJ (which I will explain later) is not available to be tuned though. And it is clearly missing from the Linux LMK.
The ADJ is the so-called priority of the app categories, with 0 being the highest (should not be killed) and 15 being the lowest (should be the first to be killed). I haven't seen values other than 0-15 but I have a feeling that the numbers are arbitrary, and can be extended up to 65535, but who would want to do that since they're only 9 app categories
LMK are specified by pages, with memory = 4 * pages.
e.g.
4096 pages represents 16MB RAM (16 *1.024)
From the Linux LMK source code, the parameters can be adjusted as such:
Code:
* For example, write "0,8" to /sys/module/lowmemorykiller/parameters/adj and
* "1024,4096" to /sys/module/lowmemorykiller/parameters/minfree to kill processes
* with a oom_adj value of 8 or higher when the free memory drops below 4096 pages
* and kill processes with a oom_adj value of 0 or higher when the free memory
* drops below 1024 pages.
Take note that the oom_adj so far takes only 6 values (although I am confident that it can take more than that). Haven't experimented enough to validate that, but I shouldn't need to as assigning different oom_adj should be a given for all the app categories. This is to make the LMK intelligent enough to determine which app to kill first rather than grouping different app categories into one priority, in which a lot of popular LMK settings that devs provide do that.
And for the Dalvik LMK, my understanding is that it will remove all dalvik caches of <INSERT_APP_CATEGORY_ADJ> or higher if free RAM/pages reaches the level specified.
Hence, two things are learned here:
- Linux LMK will kill an app category of that particular RAM level or higher if free RAM drops below a certain level e.g.
Hidden_APP has a minfree of 16384 pages. If pages fall below this level, Empty_App will be killed first, free pages is recalculated, then if RAM is still below that, Hidden_Apps will be killed next.
- Dalvik LMK will clear all dalvik caches in RAM of all the App category if free RAM reaches the level e.g.
SECONDARY_SERVER_MEM is 8192 pages. When RAM reaches 8192 pages, all SECONDARY_SERVER caches are cleared.
Best Practices
Now, how do we tweak both so that it is efficient in doing its job, i.e. LMK.
From my experiments I can safely say that:
- Below 60MB free RAM you will feel lag 5-10% of the time.
- Below 52MB free RAM you will feel lag 5-15% of the time.
- Below 46MB free RAM you will feel lag 15-20% of the time.
- Below 42MB free RAM you will feel lag 20-50% of the time.
- Below 36MB free RAM you will feel lag 100% of the time.
The experience of lag increases exponentially from 60 down to 36MB. Hence, the LMK should be tweaked against that.
Also, it is considered best practices to group oom_adj settings of Linux LMK with the Dalvik LMK as when free RAM drops to that level, both the Linux LMK and Dalvik LMK will work together to free up at both sides.
The 6 oom_adj that can be tuned for Linux minfree also need to be tuned correctly, failing this would make the LMK too aggressive/under aggressive in mid levels (even if your EMPTY_APP is aggressive, if your oom_adj is selected incorrectly, you'll get to a point where memory leak occurs).
Talon LMK - This is obsolete, TalonDev is using ThunderBolt! LMK from 0.5.x
Take for example, the Talon default LMK setting where:
LMK_ADJ is 0,1,2,3,7,15.
It's EMPTY_APP is 15
It's HIDDEN_APP is 5
It's HEAVYWEIGHT, SECONDARY and BACKUP is 3
Last three values of MINFREE is 4096, 9216, 12288.
Where's the issue here? If you noticed it congrats.
If you didn't, here's the issue:
The minfree 12288 is tied to EMPTY_APP which has a OOM_ADJ of 15.
The minfree 9216 is tied to nothing but has a OOM_ADJ of 7
The minfree 4096 is tied to HEAVYWEIGHT,SECONDARY and BACKUP with the OOM_ADJ of 3.
The LMK_ADJ is important as it represents the checkpoints where the Linux kernel will check for minfrees and kill them. Here's a situation:
Memory is below 12288 pages, OOM is triggered and Linux looks for EMPTY_APP to kill and kills all of them. Job done. Memory is back to >12288 pages.
After a few days, memory is below 12288 pages again, OOM kills all EMPTY_APP, but free memory is still lower than that. It sees if the next checkpoint is fulfilled, which is 9216 pages, and yes memory is still below 9216 pages. It looks for an OOM_ADJ of 7 and above to kill, but fails to find any as 7 and above OOM_ADJ only consists of EMPTY_APP.
Here comes the quandary. The next checkpoint is 4096. We're not ever going to reach that as you'll need memory to fall below 16MB, and at that time it'll be super sluggish to even support GUI. Hence BAM, memory leak or lag or something.
Supercharger LMK 512HP - Multitasking
Let me just paste this to avoid a lot of retyping
Code:
FOREGROUND_APP_ADJ=0
VISIBLE_APP_ADJ=4
PERCEPTIBLE_APP_ADJ=2
HEAVY_WEIGHT_APP_ADJ=5
SECONDARY_SERVER_ADJ=6
BACKUP_APP_ADJ=7
HOME_APP_ADJ=1
HIDDEN_APP_MIN_ADJ=8
EMPTY_APP_ADJ=15
FOREGROUND_APP_MEM=1536
VISIBLE_APP_MEM=3072
PERCEPTIBLE_APP_MEM=1024
HEAVY_WEIGHT_APP_MEM=10240
SECONDARY_SERVER_MEM=10240
BACKUP_APP_MEM=15360
HOME_APP_MEM=1024
HIDDEN_APP_MEM=15360
EMPTY_APP_MEM=25600
LMK_ADJ="0,4,6,8,14,15"
LMK_MINFREE="1536,3072,10240,15360,20480,25600"
A few things here:
It's EMPTY_APP and HIDDEN_APP is too large, 80 and 100MB respectively, but the 80MB is a false flag hence it doesn't really matter if you tweak that and you can get a very minor degree of multitasking because of this false flag. I'll get to that in a moment.
It's PERCEPTIBLE_APP has higher OOM_ADJ than VISIBLE_APP (odd selection, but RAM rarely goes below 40MB before it starts lagging like hell) People would reboot their phones when it gets to that, but it rarely gets to that for this configuration as LMK is pretty aggressive here.
Now, it's LMK_ADJ is "0,4,6,8,14,15"
15 is EMPTY_APP,
14 is nothing
8 is HIDDEN_APP - it looks like it'll only kill HIDDEN_APPs when it reaches 60MB, with HIDDEN_APPs being the heavy hitter of apps as most apps will end up in this category. Hence why multitasking is still doable (to an extent) but an aggressive LMK means taking a call or switching to a browser will kill your game (losing saves). Inconvenient, but it works for some people who haven't played a long game without saving and never took a call in between. Imagine the horror LOL!
Juwe's RAM Script
Code:
if [ -e /sys/module/lowmemorykiller/parameters/adj ]; then
echo "0,1,2,4,6,15" > /sys/module/lowmemorykiller/parameters/adj
fi
if [ -e /sys/module/lowmemorykiller/parameters/minfree ]; then
echo "2560,4096,5632,10240,11776,14848" > /sys/module/lowmemorykiller/parameters/minfree
fi
No OOM_ADJ are given, hence the stock ones are in effect here.
From init.rc, this is the stock 2.3.5 OOM_ADJ.
FOREGROUND 0
VISIBLE, PERCEPTIBLE 1
HEAVY_WEIGHT, SECONDARY_SERVER, BACKUP 2
HOME 4
HIDDEN_APP 7
EMPTY_APP 15
This one also suffers from the false flag of the 2nd last minfree not being used. It will start killing HIDDEN_APP at 40MB, which is usable (you'll only get lags 15-20% of the time, and most probably after long usage). Other than that, it's pretty stable.
Misc
There are a few known scripts that tweaks the LMK as well as the app priority (yes, the priority of the process categories can be changed too by using setprop.)
Supercharger is one of them:
http://forum.xda-developers.com/showthread.php?t=991276
Take note that Talon doesn't permit Supercharger to be used (it will override it when booting through boot scripts) as Talon itself uses its own LMK settings that are optimized for ZRAM/Swap.
Kickasskernel script is also included in Supercharger. The script will clash with Zach's script in ThunderBolt! too as both of them tries to set different values for the same settings.
Note that Zach's script doesn't tweak LMK settings as i9000 Talon doesn't permit it. With that, if you're using Zach's scripts, you would need to find other ways of tweaking the LMK. Using Auto Memory Manager is good. You can download Supercharger, look at the presets and apply it using Auto Memory Manager. The preset is the only thing important here
LMK App Category Priorities
Using setprop, there is a way to set the priorities of the category. Since, app categories are different, I can only offer one recommendation: The HOME_APP_ADJ.
This prop setting sets how high of a priority your launcher is. A 1 or 2 is sufficient to make sure your launcher doesn't get killed. However, do you really want it to be on 1 or 2. Based on the previous article (LMK/OOM), you would know that OOM priorities are comparatively based on one another. Setting it 1 or 2 in stock 2.3.5 values would basically override PERCEPTIBLE/VISIBLE app priorities. Do you want your launcher to be killed only after your keyboard/music player to be killed? You decide
Journaling/Barriers
This has been a touchy subject here in XDA for most people who debate about it. Most recently, the SAS/Acid tweaks included a way to disable journaling on these partitions:
/system (System is read only, it's safe to remove journaling. However, you will not see speed increase by removing it as you're not writing onto /system 99.99% of the time unless you're using Titanium backup to remove system apps or copying init.d scripts to it)
/cache (Cache can be rebuilt on the fly. Data corruption on it is not game breaking)
/data (All of your data on your phone is here. Removing journaling can risk data corruption. Read more below)
On whether we need journaling or not, I will pose this situation:
Journaling is required to maintain data consistency in events that could lead to data corruption. Data could get corrupted in a number of situations:
- Misbehaving app that constantly writes without syncing/committing data to the disk
- Power loss due to forced reboots or bootloops when data is partially written/committed into disk
Android will normally buffer its writes before committing to the disk. This way, data stays in the RAM before data would only get written into the disk after a period of time. Take note that we as Android enthusiasts seek ways to better our phone to get the most out of it through experimental and sometimes wacky ways. Take OC/UV for example. When we OCed or UVed too much, sometimes our phone will get deadsleep or constant reboots. This leads to corruption if journaling is not turned on. However, we can get corruption too even with journaling but the risk is smaller.
Mount options too, play a role in whether journaling can be safe/unsafe. Most lagfixes will place a "barrier=1" in the /data to make sure that journals are actually committed before an out-of-order write is done. With that, the risk is greatly reduced at the cost of performance. You could experiment by removing barriers and see if it works for you. No guarantees on data if you remove barriers! More info on ext4/barriers:
http://lwn.net/Articles/283161/
http://kernelnewbies.org/Ext4#head-25c0a1275a571f7332fa196d4437c38e79f39f63
If you do disable journaling, make sure to manually run an fsck on your disk. Make sure the fsck binaries are in your binary paths (echo $PATH to show your binary paths). Don't ask how to do this if you're unable to do fsck. It is your own risk that you are disabling journaling/barriers.
One experiencing I would like to relate is this:
I've personally disabled ext4 journaling on my device. After experiencing a sudden power loss (forced reboot), my apps started FCing left and right and device became unstable that it sometimes rebooted when the device is sleeping. In other words, it's a lost case. Reflashing was the only way to restore my phone. This happened 1 day after disabling my journals.
Also note that ext2 is basically an ext4 without journaling. However, ext4 is updated all the time while ext2 source is not updated for quite some time (2 or 3 years?). Most optimizations are already in ext4 while ext2 even without journaling is slower than ext4 without journaling. A comparison of performance:
No journal:
ext4 > ext2
With Journal:
ext2 (can't do journaling in ext2) > ext4 > ext3
Update: Attached is a JournalingOn.zip for i9000 only that you can flash to rejournal your partitions (Only for i9000 devices!). Also attached is Acid Tweaksv6 - Removed Useless Stuff that will disable journaling only for /system and /cache, if anyone is interested. You will need to rejournal first before unjournaling. To be sure that /system and /cache has no journaling and your mounts are applied correctly, type "mount" without the quotes.
You'll be able to see that /data will have:
Code:
barrier=1, data=ordered
/system and /cache will have:
Code:
barrier=0, data=writeback
To be really sure, you can check by using tune2fs included in both the JournalingOn.zip and Acid Tweaksv6 - Removed Useless Stuff.zip.
Copy the tune2fs to /tmp and set rwxr-xr-x as its permissions using Root Explorer/File Expert
Code:
/tmp/tune2fs -l /dev/block/mmcblk0p2 | grep features
You can substitute the /dev/block/mmcblk0p2 (which is /data) to /dev/block/stl10 (/dbdata) /dev/block/stl9 (/system) /dev/block/stl11 (/cache) to check each partition.
Continued in Post #2 - Lack of space
Still posting ...
Undervolting/Overclocking
This is also another touchy subject, but more towards people complaining about phones that get forced reboots or bootloops. OCing is done to get better performance on your android. WARNING: Ocing will cause more wear and tear on your CPU and will reduce your CPU lifetime. How much it will reduce it? No one knows. It depends on how sturdy your CPU is, and that cannot be measured by any means. It could be reducing months out of year e.g. 6months out of 3 years? No one actually knows.
UVing on the other hand reduces battery drain especially when you UV your 100MHz as 90% of the time in normal usage, your CPU will idle at 100MHz. How much can it save? No one actually did a benchmark on it. Save to say, UVing does not harm your CPU, it will actually extend the life of it by reducing wear and tear Oh, bootloops do corrupt data if you're without journaling
Always remember to clear your UV/OC settings before flashing another kernel!
Take note that the ability to OC/UV depend on your phone. Some SGS can UV better or OC better and some might not.
On how to UV:
1. You can UV more on the lower frequencies than the higher frequencies.
2. Do a UV for
100MHz with (950-150=800) - Take note that 950 is the base value for the voltage. Some kernels might have changed this value to 925 or 900. This is why clearing your OC/UV settings is important as the OC/UV settings will only load the voltage difference, which in this case -150. If the base voltage is 900 and you do a -150 for it, you'll end up with 750 which might cause bootloops.
3. Limit the max frequency to 100MHz. WARNING: Doing this will significantly slow down your system to a crawl, but it is still running. This is done only to test the stability of your system. You can skip this if you want, but you would need to find a way to stress test your CPU at the 100MHz value (and other values as well by limiting the max frequency). Failing in testing this will result in deadsleeping (as your phone is in 100MHz when you sleep and if you undervolt too much, you deadsleep)
3. Test with RockPlayer (or MX Video Player) and play an RMVB file using software decoding for 10mins. If your settings is unstable, Android will freeze up and reboot. (this is just an example, but I find that this way stresses the CPU to the max. I fail to find another way of stressing it as even playing NFS is stable on some voltages but fails when I play rockplayer. Hence rockplayer is the best way so far)
4. After doing this, remove the limit and move on with the next few frequencies.
The rest of the voltages, just rinse and repeat.
200MHz with (950-100 = 850)
400MHz with (1050-75 = 975)
800MHz with (1200-25= 1175)
1000MHz with (no change)
You can experiment with lesser values than these if you get freezes/bootloops as this is by no means a one size fits all value. Some might think this UV is overly aggressive or some might think that this UV value is too underwhelming.
Mine if you're interested:
Code:
100MHz = -200 (950 base)
200MHz = -150 (950 base)
400MHz = -125 (1050 base)
800MHz = -75 (1200 base)
1000MHz = -50 (1275 base)
As for OCing, you would do the same to test the stability. Just that you will be enabling a higher clock for your L0 frequency (L0 is the highest frequency). Some kernels might add more frequency states i.e. from the 5 states above to 6. You can try using Tegrak OC app too. It all depends on your own experimentation.
Memory Leaks
If you found out that your Android is laggy after sometime and a reboot will make it faster, then you're experiencing memory leaks. "free" is a command to show your currently free memory. It will not necessarily be the same value as your phone's free memory (as it takes into account swaps and buffers)
1. Go to terminal emulator (or download the script at Post #3)
Code:
su <ENTER>
free <ENTER>
This will show your buffers and free memory.
My example:
Code:
total used free shared buffers
Mem: 348548 340280 8268 0 1320
Swap: 163836 47292 116544
Total: 512384 387572 124812
Then do this:
Code:
sync <ENTER>
echo 3 > /proc/sys/vm/drop_caches <ENTER>
Type:
Code:
free<ENTER>
And you'll realize that the buffers will reduce and free memory will increase, like this example:
Code:
total used free shared buffers
Mem: 348548 305212 43336 0 88
Swap: 163836 46804 117032
Total: 512384 352016 160368
That means it's done.
Click to expand...
Click to collapse
If this solves the laggginess problem, it's due to vfs_cache_pressure being too low and dirty_ratio being too high. However, in most cases, you don't have to do this as Linux will manage to clear the RAM in a timely manner as long as there are no apps that constantly hold on to the memory (not releasing it). Hence, only do this when necessary (lagginess and such).
With that, I've updated the system_tweak yet again, with this knowledge by making the VM settings more aggressive, which will trigger the memory leak faster. I made this so, as there's a workaround to clear the caches when it gets full.
Having a low vfs_cache_pressure and a high dirty_ratio will save battery and make your device perform faster at the cost of higher RAM usage.
You can actually automate it by using Script Manager by making a script, pasting the instructions and making a widget out of it. To learn more, read the android market description or the in-help guide in the app.
Some commonly asked questions about drop_caches that were answered by me from flolep:
http://forum.xda-developers.com/showpost.php?p=17727859&postcount=3428
http://forum.xda-developers.com/showpost.php?p=17731938&postcount=3446
http://forum.xda-developers.com/showpost.php?p=17736421&postcount=3450
Checking for Battery Drain/ Saving battery
Starting from Gingerbread, you can't really look for Partial wakelocks anymore to determine what is draining your battery at night. BetterBatteryStats is a way to check for that. It also can provide the process that drains your battery most of the time.
A high partial wakelock usually says that something is waking up your phone at night. With that, checking partial wakelock, you can see if these are the culprit:
-Network Location
-Location Manager Service
If so, disable Lattitude or any location check-in app that you're using. It's checking you in every few minutes that drains the battery.
Disabling Wifi/3G/HSDPA saves battery too. I find that Wifi/Edge drains the least amount of battery when idling. To switch to EDGE as opposed to HSDPA/3G.
Go to:
Code:
Settings > Wireless > Mobile Networks > Network Mode > GSM Only
Disabling "Use Wireless Networks" in location and security would save a little bit of battery too I guess. It's in:
Code:
Settings > Locations and Security > Use wireless networks
CSC
CSC is a folder that defines your APN settings and country/region specific configurations aside from Language/Time that is configured in build.prop(also for Samsung Apps).
Having to manually set your APN is annoying right?
Your product code being KOR by default is another annoying thing to deal with. How do I resolve this?
1. Get your CSC from your own country here: http://forum.xda-developers.com/showthread.php?t=948790
2. Extract from the zip file /system/csc/<folder> (where <folder> is your Country's CSC)
3. Use Root Explorer/File Expert to copy it to the phone's /system/csc/.
4. Use nitrality to change your product code. WARNING: This will wipe all your data.
Odexed vs Deodexed
I've been using Deodexed ROMs since I first flash my custom ROMs. One gripe I have is that I seem to have lost the stock smoothness/speed that I experience when I'm using a stock ROM. I found out that deodexing actually extracts the customization files back to the APK (using APK manager) and Odexed files are actually done to optimize the speed of those files itself.
The only reason ROM chefs are using deodexed files is that it is easier to theme. There is no need to decompile the odex files and make the changes as all the theme files are in the APK themselves.
Only other reason that I know of is the overscroll glow. Odexed files can't support it. However, that is a small price to pay for speed/smoothness that I am willing to sacrifice.
Busybox
Busybox is required to perform all of your superuser activities in your android phone. There are some problems associated with this when ROM developers decide to use a certain version of Busybox that are incompatible with the binaries that we use in our phones. When you found out that after changing your kernel/ROM:
- You still have superuser and Titanium Backup still works but,
- Root Explorer/File Expert can't copy files correctly to the /system after mounting rw
- Voltage Control/ Pimp My CPU/ Control Freak/ SetCPU cannot create the Svolt_scheduler file in /etc/init.d
You have a clashing busybox version issue. This can be remedied by:
- Installing the busybox installer from the market
- Installing busybox 1.17.x into /xbin
- Removing all other busybox binaries from your binary path. (This is optional if you found out that after installing busybox, everything works. Try it first before removing) To know your binary path, do an
Code:
echo $PATH
Nandroid Backup
Nandroid Backup is done in recovery (VolumeUp + Home + Power). In recovery, choose Backup And Restore. What it does is it backs up your:
- Boot
- System
- Data
- Cache
When you restore, you should preferably use Advanced restore and restore the /system and /data. All others can be built from scratch (except /boot). If you're converting from CM7 to Samsung Roms, it's best to start from scratch either ways. Restoring nandroid from any is considered dangerous and not supported by most cases. Best is to use Titanium Backup to backup as it can be restored both ways (only restore app though, restore data sometimes can lead to FCs)
Voodoo Color
Voodoo Color is also a point of contention among users. Some like it and some hate it with a passion. If your kernel does support it, here's a way to tweak it:
1. Tweak the color profiles. Can't see a difference, hence I would just pick Voodoo most of the time.
2. Tweak the gamma first to a comfortable value. I.e. drag the 3 sliders together to a comfortable gamma level. After that, try looking in the gradient (the gray slab below the slider) and try to find for neutral gray values. Adjust each slider until you find a neutral gray (i.e. left and right until you see that slab is truly gray)
3. Tweak the RGB multiplier until you have a comfortable level (3 sliders together). Then tweak it the same by referring to the gradient until you're comfortable.
4. Switch off your display for a while and look at your surroundings to reset your eyes. Then turn on your screen again and see if its too green/yellow/blue etc. Adjust the gamma/RGB multiplier until you're satisfied.
This might take a few rounds of testing, but in the end you'll be truly satisfied with it
i9000 Specific
i9000 Kernels - In Depth
You will need to have a stock odexed (why odexed? read a few passages below)ROM with root. You can easily do this by flashing stock XXJVS 2.3.5 and then flashing a custom kernel through Odin like:
- Dark Core - Uses Voodoo initramfs
- CF-root - Uses CF-root initramfs
- Semaphore - Uses CF-root initramfs
- Midnight - Uses Speedmod initramfs
- Galaxian - Uses CF-root initramfs
- Voodoo - Uses Voodoo initramfs
You can flash the custom kernel through Odin by placing the tar file into the PDA section and pressing Start. At this point, do not convert your partitions to ext4 (lagfix) unless you're using a Voodoo based initramfs as Voodoo based initramfs will convert your partitions to ext4 automatically at first boot.
After flashing a custom kernel through Odin, you will gain root. From there, you can choose to use any other custom kernels that use CWM to install or stick with the custom kernel that you've flashed through Odin like:
- TalonDev - Uses Voodoo initramfs
- TalonSH - Uses Voodoo initramfs
There are other kernels as well, but since I did not stress test them, I will not include them here. This thread lists the full repertoire of custom kernels that you can choose from:
http://forum.xda-developers.com/showthread.php?t=1196704
On which custom kernel to pick, you're advised to look at the individual threads original post as well as the last 5 pages of the thread to know about its stability and features. I pick TalonDev myself as it has compache as well as the latest upstream (kernel updates from the Linux OS as well as the official Android Open Source Project/AOSP and optimizations) patches.
At this point, you should decide if you would want to convert to ext4 (lagfix) or not. Your phone by default will reside in an RFS system. The RFS at this point in time is pretty stable and its performance can easily match ext4. Too bad Quadrant scores are bad for RFS, but who cares about Quadrant anyways. I myself choose to use ext4 as the TalonDev kernel has a lot of ext4 upstream patches that optimizes the usage of ext4 on Android.
If you're using Voodoo initramfs, you'll be converted to ext4 automatically.
If you're using CF-root initramfs, you will require the CF-root ext4 app to convert that can be found here:
http://forum.xda-developers.com/showpost.php?p=12651371&postcount=7
Note that on some kernels that use CF-root initramfs, the ext4 app will warn you that you're not on CF-root type kernels. Ignore this warning. You'll be able to convert to ext4 anyways unless you're on Galaxian. Certain Galaxian versions (since the kernels are not versioned, I can't tell which versions are not working) will bar you from converting to ext4 using the ext4 app. If that is the case, just flash CF-root, convert to ext4 and then flash back to Galaxian.
Also not that CF-root based initramfs kernels will remove your bootsounds on first boot. Don't be alarmed by this.
In terms of which lagfix is better (Voodoo vs CF-root lagfix), I can't tell at this point in time. Both are equally good. Since TalonDev uses the Voodoo initramfs, I am inclined to use the same when I am using the same kernel. That should be a good yardstick to follow.
BIGMEM/ Non-BIGMEM:
Some kernel developers released BIGMEM versions of their kernel:
-TalonDev BIGMEM
-TalonSH BIGMEM
-Semaphore bm version
The only difference a bigmem kernel can bring is more memory at the expense of 720p video recording. Playing a 720p video is still possible. Recording it is not on a bigmem kernel. Bigmem kernels normally have 13MB more than their non-bigmem counterparts. How this 13MB could affect you? It depends on the apps you use and how aggressive your LowMemoryKiller is.
XXJVT System apps
System apps can be frozen/removed to make your system faster/less battery draining.
In order to remove/freeze system apps, you would need Titanium Backup (Pro) or System Tuner.
The choice whether we want to remove or freeze the system apps should now be decided. Titanium Backup Pro and System Tuner uses a built in Android mechanism to freeze your system apps - PM disable. PM disable will actually disable the app from the system itself. It will not drain the battery or even launch itself. It is a safer way so to speak. Removing will physically delete the APK app from the system freeing up disk space. If you decide to remove, make sure to have a backup before proceeding. Even if you have a backup, restoring from Titanium Backup may sometime fail. Hence, do this only if you're sure you want to remove that certain app. Freezing an app will make it easier to defrost without any consequences.
Here are the apps that I disabled:
Code:
AngryGPS > Used to manually configure GPS. Not needed if GPS works OK.
BluetoothTest
Buddies Now
lcdtest
screencapture > Screen capture app
Daily Briefing
Days
EncryptApp
Factory Test
Gallery -> I use QuickPic as a replacement. It's faster than Gallery.
Google Partner Setup
Google Search > Google search widget
Home screen tips
HTML Viewer
Market Feedback Agent
Market Updater > I froze this to avoid being updated to Market 3.0. It currently sucks now
Mobile tracker > Mobile tracker, only if you use it, then keep it
Mobile tracker settings > Mobile tracker, only if you use it, then keep it
Perso
PhoneSetupWizard > Required if you're first booting/installing the ROM, not needed afterwards
PopupuiReceiver
Press Reader
Print via Bluetooth
RoseEuKor
Samsung Account
Samsung Keypad -> I use Swype beta instead
Self Test Mode
Service mode
SimDetachNotifier > Will notify you if you detached your sim card. Seems pointless
SNS > Only necessary to sync your contacts if you're using Social hub to do so or syncing calendar from Facebook
SNSAccount > Same as above
Social Hub
Software Update
Synchronise
TwLauncher -> Using Go Launcher Ex instead
wipeoutreceiver > Wipe if phone is stolen. Not needed if you don't use the Samsung service
WlanTest
wssyncmlnps
There are others that you can remove as well. I'm hoping to gather feedback on the other APKs that can be removed and what will be affected by its removal. Please post if you have more information
Disabling Lagfix
Most ROMs would advice you to disable lagfix before flashing a new ROM. Although this is not necessary in most cases (as all Gingerbread kernels support ext4 from the get-go) moving from one kernel to another might need it (to optimize the initramfs it is used on). Hence if you're swapping from one kernel (that has a different initramfs) to another. It is adviseable to undo the lagfix.
CF's lagfix can be undone by using the ext4 app.
Voodoo lagfix can be undone by using the Voodoo Control App.
Speedmod lagfix can be undone by using it's recovery.
Odexed ROM
You might say, hey if I'm using stock, I'll be missing extended power menu, battery percentage and all other theme mods that are not in the stock ROM. Fear not, you can get all these back including Gtalk2, CRT screen off etc here:
$omator's stock+
Gtalk2
There are odexed themes too, in which the developers would decompile the odex files into APKs, make the changes then compile the APKs into odex files again. It is more tedious than deodexed themes but odexing generally makes the phone smoother
You can find the themes here:
http://forum.xda-developers.com/forumdisplay.php?f=666
WIFI Issue on JVR/JVS/JVT
You might be facing some wifi connectivity issues on JVR/JVS/JVT ROMs i.e. it disconnects every few seconds.
This is due to some changes in the ROM that doesn't permit a "Forever" lease time from your modem. Change your lease time to an hour (or two) or if you don't have that option, remove your address reservation option from your modem (Yes, it's a modem only tweak, nothing you can do on your phone)
Memory Freak
Memory Freak is an app in TalonDev that tweaks the ZRAM/Swap and LMK settings. There's a golden rule to the ZRAM/Swap that is:
Swapping to and from the ZRAM will be slower if the amount of ZRAM is more than 50% of the total usable memory, . Hence, I set mine to 160MB (from the 340 total usable RAM). Swappiness in most Linux machines is 60. This means, the system will be inclined to swap in/out more than just using the RAM. I set mine to 50 as a personal preference so that there would be 50% chance that the system will use swap or RAM.
HiddenApp is the amount of free memory to be when LMK starts looking for hiddenapps to kill. I set mine to 52 for my multitasking preference. Experiment with it to see how much multitasking vs smoothness you're looking for. My preference will change in time. Check my signature for the most updated ones.
WIP:
This is obviously a work in progress article that will undergo changes as the Android scene improves. I will spend time to make edits and additions to the post whenever I see fit. Sometimes I will post changelogs at the end of the thread, sometimes I won't if I only make minor edits. Just be sure to check back often to see any new tips offered
Credits:
I claim no credit on these findings except for the ones that I've researched on my own. However, all of it is not of original work since I form my own opinions after reading a lot. I also don't claim that the settings I posted are the most optimal there is. It might not even work well for most of you guys. The key here is tweaking it to your own perfection as there is no setting that will cater to everyone's needs. Hence the real people who deserve credits are:
* All the devs' kernel/apps that I posted here
* XDA forum
* Google
Attention!
If you're looking for scripts, here it is:
[CWM][SCRIPTS][TWEAKS] Thunderbolt!
Script Reviews are here:
Script Reviews
GUIDE UPDATES
UPDATE 1/10/12
* Edited some sections with some new knowledge that I have. A few places, can't name them all Check them out and see if you can spot them.
UPDATE 10/19/11
* Moved around some guides and grouped them together
* Updated some guides to make them more Android centric (as opposed to i9000 centric)
UPDATE 10/18/2011 - Added an indepth article about LMK/OOM. CTRL-F for LMK.
UPDATE 10/12/2011
* Added some comments on the apps removed/frozen so that you can make an informed choice of why the apps can be removed/frozen.
* Added my analysis of most schedulers that I know of and why they are good/bad.
* Moved the scripts to ThunderBolt!: LINK
* Added a guide on the wifi issue for JVR/JVS/JVT
UPDATE 10/1/2011 - Moved the script reviews to this post LINK
UPDATE 9/23/2011:
* Removed the JournalingOn.zip and Acid Tweaksv6 - Removed Useless Stuff.zip because people might use it for devices other than the i9000. This is because in other devices such as SGS4G, the /dev/block/stl10 is /data while in i9000 it's /dbdata. The JournalingOn.zip will not enable journaling on /dbdata without some editing. Hence, I removed it as I don't have a time to put up a warning or notice, and didn't have time to edit the script inside the zip file.
* Readded an edited JournalingOn.zip and readded the Acid Tweaksv6 - Removed Useless Stuff.zip (did not change) to the first post. Only for i9000 devices!!
* Added another way to check if you're journaled or unjournaled. Check the guide below.
* Added some more explanation on common scripts I found.
* Moved the guides around due to lack of space. I added some minor changes to most of the guides. See if you can detect them
Reserved - 10 char
Reserved - 10 char again ...
Changed the name to get more attention. Surprised that no one gave a comment about it
Appreciate any comments or criticisms or correction of what I've written here.
Thanks!
Excellent Read!
Thank you very much!
carnagecjb said:
Excellent Read!
Thank you very much!
Click to expand...
Click to collapse
Glad you enjoyed it
this is actually really helpful. half the time I really didnt know what I was messing around with thanks.
Dataslycer said:
this is actually really helpful. half the time I really didnt know what I was messing around with thanks.
Click to expand...
Click to collapse
Cool Glad that you liked it.
Hope this gets more attention as I would want it to be beneficial to people who uses tweaks and to know how to use them
Nice read Mate. I used to wonder about I/O schedulers and which one to use etc. I like to add build.prop scripts and int.d scripts. This thread just helped me to understand more about it.
Thanks!
read every word, thank you very much. I'll be talking to my kernel dev about BIGRAM... on miui our 720p recording doesn't work right anyways. Thanks again for the great info.
Sent from my MIUI SCH-i500
ShyamSasi said:
Nice read Mate. I used to wonder about I/O schedulers and which one to use etc. I like to add build.prop scripts and int.d scripts. This thread just helped me to understand more about it.
Thanks!
Click to expand...
Click to collapse
Thank you for reading. Hope the knowledge comes in handy when you tweak your phone
sageDieu said:
read every word, thank you very much. I'll be talking to my kernel dev about BIGRAM... on miui our 720p recording doesn't work right anyways. Thanks again for the great info.
Sent from my MIUI SCH-i500
Click to expand...
Click to collapse
Wow, great! Hope you enjoy reading it as much as I enjoyed writing it
having trouble with your thunderbolt tweak. I downloaded the file for I9000 placed it on my phone (SD card) mounted my system and data in recovery mode then proceeded to install the thunderbolt zip file. Using script manager i found the bolt_scripts file and ticked the run at boot button. Then i hit the run button and got 14 permission denied messages asking are you root. Do i have to run this as root. Might be a dumb question but im very new to trying this. Thanks
kevnac said:
having trouble with your thunderbolt tweak. I downloaded the file for I9000 placed it on my phone (SD card) mounted my system and data in recovery mode then proceeded to install the thunderbolt zip file. Using script manager i found the bolt_scripts file and ticked the run at boot button. Then i hit the run button and got 14 permission denied messages asking are you root. Do i have to run this as root. Might be a dumb question but im very new to trying this. Thanks
Click to expand...
Click to collapse
Yes, you need to run everything as root . Scripts in init.d would already be running as root. You just need to make sure Script Manager is running as root for the remount script
Excellent thread, very well written, had a quick skim read, but I'm going to take some time tonight to absorb it all ;-).
Thank you!
zeekiz said:
Excellent thread, very well written, had a quick skim read, but I'm going to take some time tonight to absorb it all ;-).
Thank you!
Click to expand...
Click to collapse
Cool Thank you for reading such a long article
Let me know if you have any questions or if you think I made a mistake
superb thread !
testomat said:
superb thread !
Click to expand...
Click to collapse
Thank you for reading
testomat said:
superb thread !
Click to expand...
Click to collapse
+1
10 chars
rom-g said:
+1
10 chars
Click to expand...
Click to collapse
Thank you for the support
Hi there,
I am on MIUI GB (2.4.13) and I noticed the same issue on CM7, so I guess this is kernel related as I already tried to uninstall every app on my phone.
After 3 or 4 days of use, I can feel the phone waking up slower and having more lags during sweeping home pages or even typing sms.
I use setCPU with no overclock and tried smartass, ondemand and interactive governors, each time having the same issue.
I think this is related to memory usage because after a fresh boot, my phone is about 450 mb free, if I launch successivly all my apps i reach about 400 mb and the phone is still flying.
But after some days of use the available memory drops to 300 mb and lower, and each time I notice this comes with the lags.
Does anyone experience the same issue, or better, knows a way to fix it ?
I don't post really ofter so I will also thank Nexx and superkid for their work on the early stage of the Desire S developpement.
Anyone on this ?
Just to confirm that I am not alone >.<
Maybe a thread somewhere already talks about this...
A few things you could try:
1) V6 Turbocharger by Zeppelinrox HERE
If your new to this you can look here
Using SuperCharger Starter Kit
Click to expand...
Click to collapse
near the bottom of the thread.
Using script manager to open the *.pdf, set as su save then run it you will have many options including 'quick engine flush' and 'detailing' which will help your particular problem. These actions and others can run from your home screen too via widget as a short background process. Once you've ran the script with script manager, exit script manager, choose SM widget from home-screen and find the particular script you ran before. Then you can flush whenever you experience lag building up.
2) Various Cache cleaning programs such as SDMaid etc...
3) A start-up manager to limit which programs boot on start-up. System Tuner Pro is a good all-round program you can try, that is feature rich.
Good Luck
Many thanks, I'll try V6 stuff, the developper is a good seller.
I'll install it on monday and tell my feedback here after 4 or 5 days.
Hi Guys,
Let me present to you my first published app in google play. it is both free and ad-free and requires no permissions.
Trimmer allows you to trim your device NAND chip manually, which may improve your device storage speed significantly.
it supports lollipop as well and I am going to add more features to this app in future so stay tuned
In order to use this app your device should be rooted and your device NAND chip should support TRIM command.
Let me know what you think and send me your suggestions for next update.
Here is a short explanation about what this app does :
Almost all Android devices use NAND chips for data storage. Because of how NAND chips work, they may become significantly slow after some time. Trimming would help to restore NAND Chips performance to their initial value. Although Android 4.3+ supports trim natively, users have no control over this process and Android doesnt trim the storage as much as it is supposed to do, which results in slow storage performance and thus laggy devies. Trimmer enables you to trim your device manually if you feel it has become laggy.
Trimmer incorporates a well-known fstrim utility which is a part of Linux kernel tools. This utility trims memory devices which support TRIM operation.
Click to expand...
Click to collapse
Download : Google Play
If you liked my app, please dont forget to rate it 5 stars. your stars will give me motivation to spend more time developing and add more features to this app.
Sounds interesting!
Would also be nice to put trimming as startup operation in init.d, can the fstrim binary be called directly by a shell?
A question: when i connect my device to a PC as usb storage to remove files from SD card, then the freed space is not reported by system till next reboot; does trimming fix also this (reporting right free space)?
[must-ask-question]Does this app perform potentially dangerous operations on filesystems?[/must-ask-question]
Publiuss said:
Sounds interesting!
Would also be nice to put trimming as startup operation in init.d, can the fstrim binary be called directly by a shell?
A question: when i connect my device to a PC as usb storage to remove files from SD card, then the freed space is not reported by system till next reboot; does trimming fix also this (reporting right free space)?
[must-ask-question]Does this app perform potentially dangerous operations on filesystems?[/must-ask-question]
Click to expand...
Click to collapse
putting trimming in init.d is not good enough. init.d only runs at startup while device storage needs to be trimmed more than that. at least once or twice per week based on usage.
yes, you can install busybox which includes fstrim and call it from shell.
great idea,good work.thx
fifthelement said:
putting trimming in init.d is not good enough. init.d only runs at startup while device storage needs to be trimmed more than that. at least once or twice per week based on usage.
yes, you can install busybox which includes fstrim and call it from shell.
Click to expand...
Click to collapse
Ehm... I reboot quite often than twice per week...
I have questions, why I will need press trim now everytime when I open the trimmer apps? It is normal... Is not stick/set on boot. Thanks in advance.
bbivan said:
I have questions, why I will need press trim now everytime when I open the trimmer apps? It is normal... Is not stick/set on boot. Thanks in advance.
Click to expand...
Click to collapse
Trim has nothing to do with boot. last time my phone was rebooted was few months ago, while some devices need to be trimmed once or twice weekly to keep optimal performance. I may add a feature in future to schedule the Trim automatically without the need of trimming manually.
fifthelement said:
Trim has nothing to do with boot. last time my phone was rebooted was few months ago, while some devices need to be trimmed once or twice weekly to keep optimal performance. I may add a feature in future to schedule the Trim automatically without the need of trimming manually.
Click to expand...
Click to collapse
Wow, uptime warrior! :laugh:
However, rated 5* on playstore!
Publiuss said:
Wow, uptime warrior! :laugh:
However, rated 5* on playstore!
Click to expand...
Click to collapse
Thanks buddy for rating
I have a GPE HTC One m7 and I only reboot it when I want to flash something like a new ROM or Kernel. otherwise it is pretty stable and there are no reboots.
I guess your devices are unstable
significantly improved with my devices...
S3 i9300 - 4.4.4 Kitkat
S2 i9100 - 5.0.2 Lollipop
Note N7000 - 4.1.2 JB with trim kernel support
Thanks buddy...
I'm a bit puzzled about what the app actually does.
If this NAND issue really is a problem to IO performance, why isn't this taken care of by the native android system?
No disrespect to your App! You'll have to excuse my ignorance here... If it actually solves such a problem, it great!
CyanMod-X said:
significantly improved with my devices...
S3 i9300 - 4.4.4 Kitkat
S2 i9100 - 5.0.2 Lollipop
Note N7000 - 4.1.2 JB with trim kernel support
Thanks buddy...
Click to expand...
Click to collapse
You're welcome Enjoy.
Please could you add support to trim /efs and /preload?
Sent from my GT-I9100 using Tapatalk
/system needs attention on the screen layout.
RovG said:
Please could you add support to trim /efs and /preload?
Sent from my GT-I9100 using Tapatalk
Click to expand...
Click to collapse
I am not sure if trimming /efs and /preload has any effect on the performance of the device and AFAIK only some Samsung devices have these partitions.
trimming is only effective for the partitions which are written to frequently.
RovG said:
/system needs attention on the screen layout.
Click to expand...
Click to collapse
I will update the layout for small screen devices in next update.
Rated 5 stars for the great effort!
Automatic trim stopped
For a few weeks the automatic trim worked flawless and I got a notification it had done it's job.
Since a few days it stopped automatic trimming. Manually it still does it's job.
Any ideas ??
Hi ..sorry for Noob question. ... I've note n7000 rooted with Omnirom. I never tried lagfix fearing the famous brickbug issue. Can I use this app without fear?
I searched for a solution following the failure of my previous fstrim solution after updating to 5.1.1 on my Nexus 7 2012, and found your work. Thank you for very much for your work.
WarriorMax said:
Hi ..sorry for Noob question. ... I've note n7000 rooted with Omnirom. I never tried lagfix fearing the famous brickbug issue. Can I use this app without fear?
Click to expand...
Click to collapse
Same question here with the Samsung Galaxy S2 (i9100) - can I safely use this app without "superbricking" my phone sooner or later?
[E]: http://forum.xda-developers.com/and...roject-brickbug-aftermath-recovering-t2823051
Apparently it's probably to use the app if you are running a current ROM.